Knowledge-based Question Answering System For College Admission Counseling Design And Implementation

It becomes a hot issue for the candidates and their parents to get the relevant enrollment information. Commonly two ways can be severed for it:the traditional search engines or artificial means:Q (Web site online service or telephone) mode. However, we can only get the web page by the traditional search engines, which will takes users much time and great efforts to screen and obtain the required information and the exact reply. In terms of the Q mode, it will result in a heavy burden because of the manpower and material resources involved in fulfillment, even though it can help us to get the most precise response in the shortest time. In this paper, it focuses on enhancing the existing search engine capabilities, allowing the system to support the natural language answers, and giving reply timely and accurately by the knowledge-based question-answering system for college admission counseling which overcomes the current shortcomings in the consultation service.This paper studies the fulfillment the components in the system modules by applying the J2EE and the MySQL by introducing an overall design and system processes for the admission consulting answering system. It summaries conclusions for the previous research and provides prospect ion for the future work as well. The specific contents of this paper and the related research results are as followed:Firstly, Summarizing 150 couples of frequently asked questions and answers from the admission counseling on the form library (FAQ), and saving the various types of information in our college brochures in into a document library, the combination of both constitutes a system of knowledge.Secondly, In terms of the question analysis process, it has been proposed to analysis and classify the types of questions based on the rules and principles, and improved the existing FMM algorithm (maximum forward match algorithm) which can help to make use of the positive incremental matching algorithm and the questions segmentation; using TF-IDF function Calculate the weight of the word W weight. Finally, In terms of the answers retrieval process, The answer can be achieved from the document automatically by information retrieval techniques and answer extraction technology. The efficiency and accuracy have been greatly improved making use of the vector space model algorithm in the calculation of similarity between the user questions and answers from FAQ database. If the answers can not be reached from the FAQ the traditional vector space model algorithm can be devoted to searching the extract answers from text databases, while the self-feedback mechanism can automatically update the answer to the FAQ database in return to the users'answers at the same time.